Jumps and Information Flow in Financial Markets

Suzanne Lee* (Georgia Institute of Technology)

Abstract
We investigate the dynamics and predictability of stochastic jump arrivals in asset prices. Specifically, we first introduce a new two-stage semi-parametric jump predictor test to determine the informational covariates that can affect jump occurrences up to the intra-day levels. Then, we examine the jump dynamics from high-frequency transaction prices of individual stocks in the Dow Jones Industrial Average. We find that the jump comes irregularly and the size distributions are highly skewed to the left and have high excess kurtosis. Jumps occur in the morning when there is corporate-specific news release, and opening prices do not necessarily include jumps. We also find that the evidence of jump clustering in normal trading: the likelihood of future jump arrivals becomes higher when there are jumps in previous trading hours.